\[\frac{x \cdot e^{\left(y \cdot \log z + \left(t - 1\right) \cdot \log a\right) - b}}{y}\]

↓

\[\frac{x \cdot \frac{{a}^{\left(-1\right)} \cdot {1}^{\left(t + y\right)}}{e^{\left(b - y \cdot \log z\right) - t \cdot \log a}}}{y}\]

Error

The X axis uses an exponential scale — Bits error versus `x`

Target

Original	2.0
Target	11.0
Herbie	1.2

\[\begin{array}{l} \mathbf{if}\;t \lt -0.88458485041274715:\\ \;\;\;\;\frac{x \cdot \frac{{a}^{\left(t - 1\right)}}{y}}{\left(b + 1\right) - y \cdot \log z}\\ \mathbf{elif}\;t \lt 852031.22883740731:\\ \;\;\;\;\frac{\frac{x}{y} \cdot {a}^{\left(t - 1\right)}}{e^{b - \log z \cdot y}}\\ \mathbf{else}:\\ \;\;\;\;\frac{x \cdot \frac{{a}^{\left(t - 1\right)}}{y}}{\left(b + 1\right) - y \cdot \log z}\\ \end{array}\]

Derivation

Initial program 2.0
\[\frac{x \cdot e^{\left(y \cdot \log z + \left(t - 1\right) \cdot \log a\right) - b}}{y}\]
Taylor expanded around inf 2.0
\[\leadsto \frac{x \cdot \color{blue}{e^{\left(\log 1 \cdot t + \left(\log 1 \cdot y + 1 \cdot \log \left(\frac{1}{a}\right)\right)\right) - \left(1 \cdot \log 1 + \left(b + \left(\log \left(\frac{1}{z}\right) \cdot y + \log \left(\frac{1}{a}\right) \cdot t\right)\right)\right)}}}{y}\]
Simplified1.2
\[\leadsto \frac{x \cdot \color{blue}{\frac{{a}^{\left(-1\right)} \cdot {1}^{\left(t + y\right)}}{e^{\left(b - y \cdot \log z\right) - t \cdot \log a}}}}{y}\]
Final simplification1.2
\[\leadsto \frac{x \cdot \frac{{a}^{\left(-1\right)} \cdot {1}^{\left(t + y\right)}}{e^{\left(b - y \cdot \log z\right) - t \cdot \log a}}}{y}\]

Reproduce

herbie shell --seed 2020184 
(FPCore (x y z t a b)
  :name "Numeric.SpecFunctions:incompleteBetaWorker from math-functions-0.1.5.2, A"
  :precision binary64

  :herbie-target
  (if (< t -0.8845848504127471) (/ (* x (/ (pow a (- t 1.0)) y)) (- (+ b 1.0) (* y (log z)))) (if (< t 852031.2288374073) (/ (* (/ x y) (pow a (- t 1.0))) (exp (- b (* (log z) y)))) (/ (* x (/ (pow a (- t 1.0)) y)) (- (+ b 1.0) (* y (log z))))))

  (/ (* x (exp (- (+ (* y (log z)) (* (- t 1.0) (log a))) b))) y))